Abstract
Knowledge of global solar radiation is required in modeling, performance investigation, and planning of solar energy projects. In this work, anewcorrelation is proposed to estimate the daily cumulated solar flux in the gulf of Tunis. This correlation is an adequate method to predict cumulated solar flux in this region. Furthermore, several models were applied to predict global solar radiation on horizontal planes from sun covering-rate. These models were derived from the original Angstrom linear regression and modified functions. Six kinds of regression have been adopted, namely linear, quadratic, cubic degree, fourth degree, exponential, and logarithmic. The models were tested for the gulf of Tunis (36 degrees 43'04"N latitude and 10 degrees 25'41"E longitude), using about 29,388 collected data during 2008 basing on 10 min time step. It is found that the sun covering-rate derived estimates basing on the simple linear Angstrom correlation are confined to within 6% of the measured values, giving reliability to the approval of the determined regression coefficients of the long-term desired predictions in the gulf of Tunis. Then, predicted values of global solar radiation on horizontal surfaces by the adopted models were compared to the measured values on the basis of statistical R-squared analysis for yearly and monthly scales. The results show that forecasting global solar radiation by the tested approaches, basing on sun covering-rate, has proved to be sufficiently reliable but the polynomial fourth degree model showed the best estimation of global solar radiation. (C) 2012 Elsevier B.V. All rights reserved.